import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data

mnist=input_data.read_data_sets('MNIST_data',one_hot=True)

#每個批次的大小
batch_size=100
#計算一共有多少個批次
n_batch=mnist.train.num_examples//batch_size

#初始化權值
def weirht_variable(shape):
    initial=tf.truncated_normal(shape,stddev=0.1)
    return tf.Variable(initial)

#初始化偏置
def bias_variable(shape):
    initial=tf.constant(0.1,shape=shape)
    return tf.Variable(initial)

#卷積層
def conv2d(x,W):
    #x input tensor of shape '[batch, in_height, in_width, in_channels]'
    #W filter / kernel tensor of shape [fileter_height,fileter_width,in_channels,out_channels]
    #strides[0]=strides[3]=1  strides[1]代表x方向的步長，strides[2]代表y方向的步長
    #padding:A 'string' from :'SAME','VALID'
    return tf.nn.conv2d(x,W,strides=[1,1,1,1],padding='SAME')

#池化層
def max_pool_2x2(x):
    #ksize [1,x,y,1]
    return tf.nn.max_pool(x,ksize=[1,2,2,1],strides=[1,2,2,1],padding='SAME')

#定義兩個placeholder
x=tf.placeholder(tf.float32,[None,784]) #28*28
y=tf.placeholder(tf.float32,[None,10])

#改變x的格式轉換為4D的向量[batch,in_height,in_width,in_channels]
x_image=tf.reshape(x,[-1,28,28,1])

#初始化第一個卷積層的權值和偏置
W_conv1=weirht_variable([5,5,1,32]) #5*5的取樣視窗，32個卷積核從1個平面抽取特徵
b_conv1=bias_variable([32]) #每一個卷積核對應一個偏置值

#把x_image和權值向量進行卷積，再加上偏置值，然後應用於relu啟用函式
h_conv1=tf.nn.relu(conv2d(x_image,W_conv1)+b_conv1)
h_pool1=max_pool_2x2(h_conv1) #進行max_pooling

#初始化第二個卷積層的權值和偏置
W_conv2=weirht_variable([5,5,32,64]) #5*5取樣視窗，64個卷積核從32個平面抽取特徵
b_conv2=bias_variable([64]) #每一個卷積核對應一個偏置值

#把h_pool1和權值向量進行卷積，再加上偏置值，然後應用於relu啟用函式
h_conv2=tf.nn.relu(conv2d(h_pool1,W_conv2)+b_conv2)
h_pool2=max_pool_2x2(h_conv2) #進行max_pooling

#28*28的圖片第一次卷積後還是28*28，第一次池化後變成14*14
#第二次卷積後為14*14，第二次池化後變成7*7
#進行上面的操作後得到64張7*7的平面

#初始化第一個全連線層的權值
W_fc1=weirht_variable([7*7*64,1024])
b_fc1=bias_variable([1024])

#把池化層2的輸出扁平化為1維
h_pool2_flat=tf.reshape(h_pool2,[-1,7*7*64])
#求第一個全連線的輸出
h_fc1=tf.nn.relu(tf.matmul(h_pool2_flat,W_fc1)+b_fc1)

#keep_prob用來表示神經元輸出概率
keep_prob=tf.placeholder(tf.float32)
h_gc1_drop=tf.nn.dropout(h_fc1,keep_prob)

#初始化第二個全連線層
W_fc2=weirht_variable([1024,10])
b_fc2=bias_variable([10])

#計算輸出
prediction=tf.nn.softmax(tf.matmul(h_fc1,W_fc2)+b_fc2)

#交叉熵代價函式
cross_entropy=tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=y,logits=prediction))
#使用AdamOptimizer進行優化
train_step=tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)
#結果儲存在一個布林列表中
correct_prediction=tf.equal(tf.argmax(prediction,1),tf.argmax(y,1))
#求準確率
accuracy=tf.reduce_mean(tf.cast(correct_prediction,tf.float32))

with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    for epoch in range(21):
        for batch in range(n_batch):
            batch_xs,batch_ys=mnist.train.next_batch(batch_size)
            sess.run(train_step,feed_dict={x:batch_xs,y:batch_ys,keep_prob:0.7})

        acc=sess.run(accuracy,feed_dict={x:mnist.test.images,y:mnist.test.labels,keep_prob:1.0})
        print("Iter "+str(epoch)+",Testing Accuracy="+str(acc))
 

Iter 0,Testing Accuracy=0.8593
Iter 1,Testing Accuracy=0.8752
Iter 2,Testing Accuracy=0.8771
Iter 3,Testing Accuracy=0.8826
Iter 4,Testing Accuracy=0.8821
Iter 5,Testing Accuracy=0.8835
Iter 6,Testing Accuracy=0.8825
Iter 7,Testing Accuracy=0.8853
Iter 8,Testing Accuracy=0.8881
Iter 9,Testing Accuracy=0.888
Iter 10,Testing Accuracy=0.8872
Iter 11,Testing Accuracy=0.8878
Iter 12,Testing Accuracy=0.985
Iter 13,Testing Accuracy=0.9897
Iter 14,Testing Accuracy=0.9887
Iter 15,Testing Accuracy=0.9876
Iter 16,Testing Accuracy=0.991
Iter 17,Testing Accuracy=0.9912
Iter 18,Testing Accuracy=0.991
Iter 19,Testing Accuracy=0.9888
Iter 20,Testing Accuracy=0.9908